Maternal mental health problems can negatively impact children's physiological stress regulation. Yet, little is known of their long‐term effects, especially related to the timing of maternal symptoms. We examined how maternal mental health problems during pregnancy versus in the early postpartum period predict children's cortisol levels and diurnal patterns at 10–12 years. Participants were a selection (N = 102) of an original sample of 805 Finnish families, who were followed from the second trimester of pregnancy (T1) to child's age of 2 months (T2) and 12 months (T3), and again at child's age of 10–12 years (T4). Based on the timing of psychological distress and depressive symptoms (T1–T3), the mothers could be assigned to three distinct mental health trajectory groups: mothers with prenatal mental health problems (n = 15), mothers with early postpartum mental health problems (n = 15) and mothers without mental health problems (n = 72). Children's cortisol (T4) was measured by saliva samples through five within‐1‐day assessments. The results show that maternal prenatal mental health problems predicted a relatively steep increase of child cortisol from awakening to 1 h later, indicating an intensified cortisol awakening response (CAR). Mothers' early postpartum mental health problems instead predicted a reduced CAR. Both maternal prenatal and postnatal mental health problems thus predicted children's later stress regulation, but in unique ways. We discuss the specific roles of direct biochemical effects during pregnancy and postpartum mother–infant interaction quality as modifiers of the hypothalamic–pituitary–adrenal system. Copyright © 2015 John Wiley & Sons, Ltd.